Create INDSCAL Carroll & Wish example...

Creates eight Dissimilarity objects that bear names "1" ... "8".

These objects contain the interpoint distances for a twodimensional 3×3 Configuration of points, labelled A, B, C, ... I. All Dissimilarity objects are based on the following underlying configuration.

The eight sources weigh this configuration in the following manner:

For each source, the distances were subjected to the transformation:

dissimilarityij = distanceij + noiseRange · u,

where u is a uniform random variable between 0 and 1.

Now you can do the following for example:

Select all the Dissimilarity objects and choose To Distance....
Uncheck scale (add "additive constant").
Select all the Distance objects and choose To Configuration (indscal)....
and an INDSCAL analysis will be performed. In order to reproduce the saliences, you have to uncheck the "Normalize scalar products" option.

This example was adapted from Carroll & Wish (1974).

Links to this page

© djmw, December 1, 1997